Converting an old pretrained model to use with the current version of deepspeech

[JRMeyer]

>>> ystepanoff
[February 26, 2021, 7:59pm]

Dear all,

I have a language model trained with an old version of DeepSpeech (I
think it was 0.1.0 or something, so I have all the files like
output_graph.pb, lm_binary, and trie). Is it possible to convert it so
that it can be used with the current version of DeepSpeech (i.e. I
somehow want to obtain a pbmm file which can be used with DeepSpeech
0.9.3)?

Thank you very much for your help and all the good stuff!

[This is an archived TTS discussion thread from discourse.mozilla.org/t/converting-an-old-pretrained-model-to-use-with-the-current-version-of-deepspeech]

[JRMeyer]

>>> lissyx
[February 26, 2021, 10:54pm]

> Is it possible to convert it so that it can be used with the current
> version of DeepSpeech

No, you need to retrain from scratch, sorry

[Archived Post]